One chart is burned into my mind from election night.
The chart above — from the New York Times election night coverage — is a gauge chart, a display that information visualization practitioners rail against frequently. This is partly because they waste a lot of space (and more at the Tableau Blog too) If designers choose to represent data with a gauge, they spend a considerable amount of screen real estate to represent a single datapoint. Infovis practitioners and researchers love economy. Edward Tufte calls this the 'chart to junk ratio'... and gauges have a lot of junk.
But the design team at the NYT also chose to "wiggle" the data, using a jitter algorithm that randomly repositioned the needle. Or at least semi-randomly repositions it.
There went all of my finger nails.
But now, in the aftermath of the election, I just got to read the rationale for choosing jitter, which I hadn't considered too deeply, thinking it a cruel trick and not a considered choice. Gregor Aisch wrote up a nice little piece that outlines their thinking. Some highlights:
- Jitter is good for showing visitors the data is "live" and that any changes would not require a page refresh
- The jitter had a purpose - it captured the uncertainty of the forecast. As the confidence in the forecast went up, the jitter went down. I recall that the jitter was very pronounced at 8 and 9 PM and shrunk to almost no wiggle by 11 or 11:30 PM. I noticed this on the night of the election, I think. Though it's really subtle and now I'm not sure if I did remember it on election night or I'm just projecting some rationality of my reading of the gauge.
- The designers added noise to the jitter. Which I'd argue makes it more "fun" and dynamic. You could imagine jitter that just monotonously and smoothly swung between the two end points of the forecast. This part seems more contentious to me, but if your goal is eyeballs glued to the site, then this was a good choice.